JP2009054843A - Device, method and program for process abnormality detection - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a process abnormality detection device capable of appropriately deciding whether there is an abnormality or not in states relating to wide-ranging processes. <P>SOLUTION: A process abnormality detection device includes: a process data edit section 22 that extracts a process characteristic quantity from process data stored in a process data storage section 21; a data storage section of abnormality detection factor analysis rule 26 that stores abnormality detection factor analysis rules for detecting an abnormality in a manufacturing system on the basis of the process characteristic quantity; a process result abnormality decision section 24' that decides whether there is a product abnormality or not on the basis of the process characteristic quantity in accordance with an abnormality detection factor analysis rule (PLS) requiring objective variables; and a device abnormality decision section 24" that decides whether there is an abnormality in a device state or not in accordance with an abnormality detection factor analysis rule (PCA) requiring no objective variables. The abnormality decision sections 24' and 24" are configured to operate simultaneously. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a process abnormality detection apparatus, method, and program for detecting abnormality relating to a process.

  The manufacturing process of various products including a semiconductor / liquid crystal panel must be appropriately managed in order to improve the manufacturing yield of the product or maintain a good yield.

  A semiconductor device is manufactured through a semiconductor process having 100 steps or more, and is manufactured using a plurality of complicated semiconductor manufacturing apparatuses. For this reason, there are many cases where a relationship between a parameter indicating the state of each manufacturing apparatus (process apparatus) and characteristics of a semiconductor device manufactured using each manufacturing apparatus is not clearly required. On the other hand, in the semiconductor process, there is also a demand that each process must always be strictly managed so that the yield of manufactured semiconductor devices is improved.

In order to solve such a problem, in the invention disclosed in Patent Document 1, a wide variety of process data and processing results generated during process execution are acquired, and the process data and its processing are obtained from the obtained process data by the least partial square method. Create a model of the resulting correlation. Using this model, the processing result can be predicted at the time of process execution.
JP 2004-281461 A

  In the conventional invention disclosed in Patent Document 1 and the like, since one determination unit predicts the processing result using the above model at the time of process execution, for example, the state of the product that is the execution result of the process Therefore, it is difficult to appropriately determine various states such as the state of the process device executing the process.

  The invention disclosed in Patent Document 1 is based on the premise that the process apparatus operates normally and the collected process data accurately reflects the process state in the process apparatus. Therefore, for example, when a sensor or the like mounted on the process device is out of order, the process data used for prediction is not reliable, and correct prediction cannot be performed. Similarly, even if an abnormality occurs in the device that executes the process, it cannot be predicted correctly because it differs from the situation when the model is created.

  It is an object of the present invention to provide a process abnormality detection apparatus, method, and program capable of appropriately determining the presence or absence of abnormality in various states related to processes. Another object of the present invention is to improve the reliability of the determination result of the presence or absence of product abnormality.

  An abnormality detection apparatus according to the present invention is (1) an abnormality detection apparatus that detects a process abnormality based on process data obtained at the time of process execution in a manufacturing system including one or a plurality of manufacturing apparatuses, Process data storage means for storing process data, process data editing means for extracting process feature values from the process data stored in the process data storage means, and an abnormality for detecting an abnormality in the manufacturing system from the process feature values An abnormality detection rule data storage means for storing a detection rule; and a plurality of abnormality determination means for determining presence / absence of an abnormality related to the process from the process feature amount by the abnormality detection rule, the plurality of abnormality determination means Are configured to be executed simultaneously, and the plurality of abnormality determination means At least one of the plurality of abnormality determination means is determined by an abnormality detection rule that does not require an objective variable. The presence or absence of abnormality was determined.

  In the embodiment, the plurality of abnormality determination units correspond to the process result abnormality determination unit 24 ′ and the apparatus abnormality determination unit 24 ″. In the embodiment, the number of abnormality determination units is two, but may be three or more. Abnormality may be detected for each unit target product.

  (2) An abnormality detection rule that requires an objective variable can be used to determine whether there is an abnormality in a product obtained by executing a process.

  (3) An abnormality detection rule that requires an objective variable may include a rule that uses a partial least square regression model (PLS) as a prediction model of a process execution result.

  (4) An abnormality detection rule that does not require an objective variable can be used to determine whether there is an abnormality in the manufacturing apparatus for executing the process.

(5) An abnormality detection rule that does not require an objective variable may include a rule using a Q statistic and / or T ² statistic by principal component analysis.

  (6) A process abnormality detection method according to the present invention is a method in a process abnormality detection apparatus that detects a process abnormality based on process data obtained during process execution in a manufacturing system including one or a plurality of manufacturing apparatuses. A step of acquiring time series process data and storing it in the process data storage means; a step of extracting process feature values from the process data stored in the process data storage means; and the process from the extracted process feature values An abnormality determination step for determining the presence or absence of an abnormality related to an abnormality, and the abnormality determination step does not require a process for determining the presence or absence of an abnormality according to an abnormality detection rule that requires an objective variable and an objective variable The process of determining the presence or absence of an abnormality according to the abnormality detection rule It was to be executed.

  (7) Further, the program according to the present invention includes a process data editing means for extracting a process feature amount from time-series process data stored in the process data storage means, and an abnormality detection rule requiring an objective variable. By means of an abnormality determination means for determining the presence or absence of an abnormality related to the process from the process feature value, an abnormality detection rule that is executed simultaneously with the abnormality determination means and does not require an objective variable, the abnormality of the process related value is detected from the process feature value. It was made to function as another abnormality determination means for determining the presence or absence.

  The abnormality detection rule data storage means corresponds to the abnormality detection factor analysis rule data storage unit of the embodiment. The abnormality determination means for determining the presence / absence of an abnormality by an abnormality detection rule that requires an objective variable corresponds to the process result abnormality determination unit 24 ′ of the embodiment, and determines the presence / absence of an abnormality by an abnormality detection rule that does not require an objective variable. The abnormality determination means that corresponds to the apparatus abnormality determination unit 24 ″ of the embodiment.

Note that the Q statistic and / or the T ² statistic are calculated for each unit target product, and when the Q statistic and / or the T ² statistic are determined to be abnormal continuously for the number of times designated in advance, the manufacturing apparatus It is preferable to provide means for notifying the abnormality. Moreover, the abnormality determination means can validate the presence / absence of an abnormality in the product when the manufacturing apparatus is determined to be normal based on the Q statistic and / or the T ² statistic. In this case, the abnormality determination unit may notify the product abnormality prediction when the effective prediction for the product is determined to be abnormal continuously for the number of times designated in advance. When it is determined whether there is a continuous abnormality, a single (accidental) abnormality can be eliminated. Furthermore, it is preferable to add a function for obtaining an abnormality factor in addition to the presence or absence of abnormality.

  In the present invention, the abnormality determination process performed using the abnormality detection rule that requires the objective variable and the abnormality determination process performed using the abnormality detection rule that does not require the objective variable can be performed simultaneously. As described above, by using different types of abnormality detection rules, it is possible to perform abnormality determination for various types of states. In addition, since each abnormality determination process can be executed simultaneously, it can be determined quickly.

  As an example, product abnormality determination based on predicted values and device status monitoring based on a wider range of process data can be performed simultaneously. Thus, the simultaneous monitoring from different viewpoints improves the accuracy of abnormality detection. Furthermore, by registering multiple models with different parts and executing them simultaneously and comparing the results, requirements that affect the quality of the product can be easily extracted.

  Target products manufactured by the manufacturing process include semiconductors and FPDs (flat panel displays: displays using liquid crystal, PDP, EL, FED, etc.). “Unit target product” may be a target product grasped in a normal counting unit such as one semiconductor wafer, one glass substrate, or grasped in a group unit of products such as one lot of these products. The target product may be a target product that is a unit of a product such as a region set on a large glass substrate. The output of the abnormality notification information includes various processes such as output to a display device, notification by mail transmission or the like, and saving in a storage device. The “product abnormality” corresponds to an abnormality in the predicted value state in the embodiment, and the “abnormality in the manufacturing apparatus” corresponds to an abnormality in the process state in the embodiment. An abnormality of a manufacturing apparatus (process apparatus) causes an abnormality in process data obtained from the manufacturing apparatus, such as a malfunction of an apparatus for executing the process of the apparatus or an abnormality of a sensor incorporated in the apparatus. including.

  In the present invention, the abnormality determination process performed using the abnormality detection rule that requires the objective variable and the abnormality determination process performed using the abnormality detection rule that does not require the objective variable are performed at the same time. On the other hand, it is possible to appropriately determine whether or not there is an abnormality.

  FIG. 1 shows a manufacturing system including an abnormality detection apparatus according to a preferred embodiment of the present invention. This manufacturing system includes a process device 1, an abnormality detection device 20, and an abnormality display device 2. These devices are connected to each other by an EES (Equipment Engineering System) network 3 which is a device network for exchanging process-related information more detailed than production management information at high speed. Although not shown, the EES network 3 is also connected to other process devices and inspection devices used in a stage before the process apparatus 1 and a stage after the process apparatus 1. Further, this system includes a production management system 4 including a MES (Manufacturing Execution System) and a MES network 5 that transmits production management information connected to the production management system 4. The EES network 3 and the MES network 5 are connected via a router 6. The production management system 4 existing on the MES network 5 can access each device on the EES network 3 via the router 6.

  This manufacturing system manufactures a semiconductor and a liquid crystal panel, for example, and the process apparatus 1 executes a process (such as a film forming process on a wafer) for manufacturing a semiconductor or the like. In a semiconductor manufacturing process or a liquid crystal panel manufacturing system, a predetermined number of wafers or glass substrates (hereinafter referred to as “wafers”) to be processed are set in a cassette 7 and moved in units of cassettes. Processing is performed. When one product is manufactured, predetermined processing is executed in each of the plurality of process apparatuses 1. In that case, movement between process devices is also performed in cassette units. A predetermined number of wafers mounted in the cassette 7 form the same lot.

  In the semiconductor manufacturing system of this embodiment, since it is necessary to manage each wafer, a product ID is assigned to each wafer. This product ID can be set, for example, by combining a lot ID and an identification number in the lot. That is, if the lot ID is “0408251” and the number of sheets that can be set in the lot is one digit, the product ID of the second glass substrate in the lot (the identification number in the lot is “2”) is It can be set to “04082512” with the identification number in the lot added to the digit.

  Of course, the product ID for all wafers stored in place of the lot ID or together with the lot ID is recorded in the tag 7a, and the process device 1 (process data collection device 12) is stored in the tag 7a. All product IDs may be acquired. When one wafer is set in the cassette 7, the ID recorded on the tag 7a can be used as it is as the product ID. When analyzing in lot units, it is not necessary to acquire a product ID or create a product ID based on the lot ID.

  An RF-ID (radio frequency identification) tag 7 a is attached to the cassette 7. The tag 7a is electromagnetically coupled to the RF-ID read / write head 8 connected to the process apparatus 1 and reads / writes arbitrary data without contact, and is also called a data carrier. The tag 7a stores information such as the lot ID (the lot ID that is the basis of the product ID or the product ID itself) and the shipping time of the preceding apparatus.

  The process apparatus 1 acquires the recipe ID sent from the production management system 4 via the router 6 from the MES network 5. The process apparatus 1 has a correspondence table between recipe IDs and processes to be actually performed, and executes processes according to the acquired recipe IDs. The process apparatus 1 is set with an apparatus ID for identifying each apparatus.

  The process device 1 includes a process data collection device 12. The process data collection device 12 is connected to the EES network 3. The process data collection device 12 collects process data that is information related to the state of the process device 1 in time series during a period in which the process is being executed in the process device 1 or during standby. The process data includes, for example, the voltage and current during operation of the process device 1 and the residence time from when the process device 1 that executes a certain process is delivered to when the process device 1 that executes the next process is charged. is there. Further, when the process apparatus 1 includes a plasma chamber and performs a film forming process on the wafer, there are a pressure in the plasma chamber, a gas flow rate supplied to the plasma chamber, a wafer temperature, a plasma light amount, and the like. The process device 1 includes a detection device for detecting these process data, and the output of the detection device is given to the process data collection device 12.

  The process data collection device 12 collects the delivery time of the previous stage device read from the tag 7a through the RF-ID read / write head 8 and the entry time to the process device 1 where the wafer is currently set. By taking the difference between the exit time and the entry time, the residence time from the preceding apparatus can be calculated. Further, the RF-ID read / write head 8 writes the delivery time and the like to the tag 7a when the wafer is delivered from the process apparatus 1 as necessary.

  The process data collection device 12 has a communication function. The process data collection device 12 collects all process data generated in the process device 1, associates the product ID and device ID with the collected process data, and outputs them to the EES network 3. The type of data to be collected is not limited to the above, and it does not prevent obtaining more information.

  The abnormality detection device 20 is a general personal computer from the viewpoint of hardware, and each function of the device is realized by an application program that runs on an operating system such as Windows (registered trademark). .

  FIG. 2 mainly shows the internal configuration of the abnormality detection device 20. The anomaly detection device 20 stores the process data of the process device 1 sent from the process data collection device 12 and the process data from various process data stored in the process data storage unit 21. Based on the process data editing unit 22 to be calculated, the process feature data storage unit 23 for storing the process feature data calculated by the process data editing unit 22, and the process feature data stored in the process feature data storage unit 23. An abnormality determination unit 24 that determines whether there is an abnormality, an abnormal process data storage unit 27 that stores process data for a wafer that is determined to be abnormal by the abnormality determination unit 24, and a determination result that stores the determination result of the abnormality determination unit 24 An abnormality detection factor used when the data storage unit 28 and the abnormality determination unit 24 perform determination processing An anomaly detection factor analysis rule data storage unit 26 for storing analysis rules, an anomaly detection factor analysis rule editing unit 25 for accessing the anomaly detection factor analysis rule data storage unit 26 to add / change an anomaly detection factor analysis rule, and It is equipped with. Each storage unit may be set in a storage device (database 20a) external to the abnormality detection device 20, or may be provided in an internal storage device.

  As shown in FIG. 3A, the process data stored in the process data storage unit 21 is associated with a product ID and a device ID. The process data includes date and time information (date + time) when the process data is collected in addition to various process data collected by the process data collection device 12. The process data storage unit 21 for each process device stores process data in time series according to date and time information for each product ID. FIG. 1 shows an example in which process data of one process device 1 is given to the abnormality detection device 20 and abnormality analysis is performed based on the process data. However, in the case where a product passes through a plurality of process devices. The process data obtained by the plurality of process devices may be given to the abnormality detection device 20. In that case, each of the above data is created by the number of devices.

  The process data storage unit 21 includes temporary storage means such as a ring buffer, and deletes process data (overwrites new process data) at a predetermined timing after the end of the process. The process data storage unit 21 is prepared for each process device 1 (data collection device 12), and the process data collected from the same process device 1 is stored in the process data storage unit 11 uniquely associated in time series. Is done. Of course, the process data acquired from each process apparatus 1 (data collection apparatus 12) may be mixed and stored in one process data storage unit 21. In that case, process device identification information such as a device ID is stored in association with each other.

  The process data editing unit 22 calls time-series process data stored in the process data storage unit 21 and calculates a process feature amount for each sheet. The process feature amount is not limited to, for example, a value calculated from process data values such as the peak value, sum total, and average value of the process data for the same product ID, and the process data value exceeds a set threshold. There are various things such as time.

  The process data editing unit 22 acquires the recipe ID output from the production management system 4 together with the product ID and the device ID. A recipe is a set of commands, settings, and parameters for a predetermined process apparatus, and a plurality of recipes are provided depending on processing targets, processes, and apparatuses, and are managed by the production management system 4. Each recipe is given a recipe ID. A recipe for a wafer processed by the process apparatus 1 is specified by an apparatus ID, a product ID, and a recipe ID.

  The process data editing unit 22 acquires a set of the product ID, the device ID, and the recipe ID shown in FIG. First, the process data editing unit 22 accesses the production management system (MES) 4 and searches for the corresponding recipe ID using the product ID of the wafer to be analyzed and the apparatus ID that identifies the process apparatus 1 as keys. Next, the process data editing unit 22 acquires the retrieved recipe ID from the production management system 4 directly or via the process data collection device 12. When acquiring via the process data collection device 12, the process data collection device 12 acquires the recipe ID of the process in progress from the production management system (MES) 4, and combines the device ID of the process device 1 and the process data. You may make it pass to the abnormality detection apparatus 20. FIG.

  The process data editing unit 22 combines the calculated process feature data and the acquired recipe ID using the product ID and the device ID as keys, and stores the combined data for the process feature data for the corresponding device ID. Stored in the unit 23. Therefore, the data structure of the process feature data storage unit 23 is as shown in FIG.

  The abnormality detection factor analysis rule editing unit 25 acquires a model obtained by the modeling device 14 or manual analysis, detects an abnormality of the process and the device, and analyzes an abnormality factor. An abnormality detection factor analysis rule including a factor analysis rule is defined and stored in the abnormality detection factor analysis rule data storage unit 26. As the modeling device 14, for example, a modeling device using data mining disclosed in Japanese Patent Application Laid-Open No. 2004-186445 can be used. Here, data mining is a technique for extracting rules and patterns from a large-scale database, and as its specific technique, a technique called decision tree analysis and a technique called regression tree analysis are known.

  Furthermore, the abnormality detection factor analysis rule editing unit 25 also registers abnormality notification information corresponding to the abnormality detection factor analysis rule. Thereby, as shown in FIG. 4, the data structure of the abnormality detection factor analysis rule data storage unit 26 includes an apparatus ID of each process device, a recipe ID of each process device, an abnormality detection factor analysis rule, and abnormality notification information. And a table structure in which

The abnormality notification information includes an abnormality display device 2 that displays a result determined based on the abnormality detection factor analysis rule, information that specifies an output destination such as a notification destination that notifies the determination result, and specific notification contents. . The notification destination is, for example, a mail address of a person in charge. Both the abnormality display device 2 and the notification destination may be registered, or only one of them may be registered. When a plurality of output destinations are provided, for example, it is possible to classify according to the degree of abnormality or the abnormality location obtained by the determination, and distribute according to the classification. A plurality of abnormality display devices, notification destinations, and notification contents can be designated for one classification. The anomaly detection factor analysis rule should use a combination of techniques such as multiple linear regression, PLS linear regression, decision tree, Mahalanobis distance, principal component analysis, moving principal component analysis, DISSIM, Q statistic, T ² statistic, etc. Can do.

  This abnormality detection factor analysis rule is a rule for detecting the presence / absence of an abnormality of a product and the presence / absence of an abnormality of the process apparatus itself from the process feature quantity. The anomaly detection factor analysis rule for predicting the presence or absence of product anomalies uses a rule that requires an objective variable, an anomaly judgment formula that performs arithmetic processing based on the process feature value, and a value obtained from the anomaly judgment formula ( and a determination condition for determining whether or not y) is abnormal. Also, abnormality detection can be performed by using a PLS (Partial Least Squares) method as abnormality detection. The abnormality factor analysis is to obtain abnormality factor data. The abnormality factor data includes process data or a name indicating the feature amount and contribution rate data.

  The contribution rate data is data representing how much process data and its feature amount influence the abnormality. It can be said that the greater the numerical value of the contribution rate data, the greater the degree of influence on the abnormality, that is, the higher the possibility of the cause of the abnormality. Abnormality factor data including up to the top N (for example, five) contribution rate data values of the contribution rate data calculated by the abnormality factor analysis is extracted. Based on the extracted abnormality factor data, the worker knows which process data should be checked when dealing with an abnormality detected.

In the present embodiment, the contribution rate for determining the abnormality factor data is obtained from a regression equation obtained by the PLS (Partial Last Squares) method. The regression equation obtained by this PLS method is shown below.
y = b0 + b1 * x1 + b2 * x2 + ... + b (n-1) * x (n-1) + bn * xn

  In the above equations, x1, x2,... Xn are process feature quantities, and b0, b1, b2,. b1, b2,..., bn are the weights of each process feature amount. When the value of y obtained by the above regression equation exceeds the threshold value, it is determined as abnormal.

In order to obtain the contribution ratio of each process feature amount using the PLS method, the following is performed. First, assume that the PLS predicted value when each variable (x1, x2,... Xn) shows an average value is Y. Then, how much each term contributes to the magnitude of yY, which is the difference from y obtained by substituting the actually acquired process feature quantity into each variable, is evaluated. That is, assuming that the average value of each variable is X1, X2,... Xn, the value of each term in the above equation is as follows.
b1 (x1-X1), b2 (x2-X2), ..., bn (xn-Xn)
As described above, the value of each term obtained by multiplying the difference between the average value and the actual measurement value by the coefficient is used as contribution rate data of each process feature amount. As a result of the factor analysis, it is possible to identify which process feature quantity is a problem.

The determination of the presence / absence of an abnormality in the process device (process state) uses a Q feature value and a T ² statistic that are rules having no objective variable. In other words, the control limit (normal space) is set using Principal Component Analysis (PCA), and the threshold obtained with reference to the value obtained using the model construction data (process feature data + inspection data) And Then, during operation (when an abnormality is detected), it is determined whether the process state is normal in real time (for each sheet) from the threshold value. Here, the Q feature amount and the T ² statistic are respectively obtained by the following equations.

Here, xp is the value of the p-th variable, and Xp is the average value of the p-th variable. tr is the r-th principal component score in the principal component analysis, and R is the number of adopted principal components. Then, when the obtained Q feature value or T ² statistic value exceeds a preset threshold value, it can be determined that an abnormality has occurred.

  The abnormality factor analysis can also be performed in the abnormality determination of the process apparatus using the PCA. In other words, the abnormality factor data is data representing how much process data and the process feature amount affect the abnormality when the abnormality location cannot be uniquely identified even when the abnormality is detected. The abnormality factor data includes a name indicating process data or a process feature amount and a numerical value indicating a contribution and an influence degree. It can be said that the greater the numerical value, the greater the degree of influence on the abnormality, that is, the higher the possibility of the cause of the abnormality.

  The contribution data is data representing how much process data and its feature amount influence the abnormality. It can be said that the greater the value of the contribution data, the greater the degree of influence on the abnormality, that is, the higher the possibility of the cause of the abnormality. Abnormality factor data including up to M (for example, five) contribution data values of contribution data calculated by the abnormal factor analysis is extracted. Based on the extracted abnormality factor data, the worker knows which process data should be checked when dealing with an abnormality detected. Note that the value of M may be the same as or different from that in the above-described abnormality factor analysis based on PLS.

In the present embodiment, principal component analysis is also used to identify this abnormal factor. In other words, performs principal component analysis multiple processes data (process characteristic quantity) as variables, Q statistic based on the following equation, T ² statistic, and calculates the contribution to the statistic for each variable.

  The arithmetic expression for determining the Q statistic contribution is an expression for determining the contribution of the p-th variable to the Q statistic. Since the Q statistic is the sum of the squared prediction errors of each variable, the contribution of the p-th variable can be obtained from this equation. If there is a variable whose contribution is extremely larger than that in the normal state, it can be determined that the variable is related to abnormality. The normal contribution value of each variable is obtained in advance and registered as one element of the abnormality detection factor analysis rule. Since the contribution of each variable at normal time is predicted to be different for each variable, for example, the contributions of all variables obtained based on the calculated process feature values of the workpiece to be judged are By dividing each by contribution, normalization is performed, and the normalized values are compared with each other, and the top M items having higher values can be identified as abnormal factors.

Similarly, arithmetic expression for obtaining the T ² statistic contribution is an equation for obtaining the contribution to T ² statistic of the p variables. t is a vector composed of principal component identification, and vp is an arrangement vector of the p-th row (coefficient vector related to the p-th variable) of the load matrix. Similarly to the case of the Q statistic, the contribution of each variable at the normal time is obtained, and the variable having a contribution greatly different from that at the normal time is extracted and set as an abnormal factor. As described above, as a result of the factor analysis, it is possible to identify which process feature amount is a problem.

  The factor analysis using this contribution can list a plurality of abnormal factors, although it cannot identify a specific abnormal location.

Thus, determination of the presence or absence of abnormality of the process equipment, using the Q characteristic quantity and T ² statistic, when at least one of Q characteristic quantity and T ² statistic exceeds the threshold, if there is abnormality in the processing device I can guess. However, Q feature amount, even if the T ² statistic exceeds the threshold value, since there is a possibility that indicates the value of the abnormality from exogenous other reasons other than the process device, in this embodiment, abnormality of the process equipment notification of occurrence, Q characteristic quantity and T ² statistic according is to perform when the phenomenon exceeding the threshold value has continued N times in a row.

The specific processing function of the abnormality detection factor analysis rule editing unit 25 related to PLS is to execute the flowchart shown in FIG. First, the process data for rule construction already collected and the inspection result data including normal / abnormal are analyzed by the PLS method to obtain a prediction model (S41). Then calculated statistic Q, the ^{T 2} from the data (S42). Then, register the prediction model and statistic Q, each threshold value for T ² and abnormality determination as rules together with the recipe ID (S43). The processing steps S41 and S42 may be executed by the modeling device 14.

  The abnormality determination unit 24 includes a process result abnormality determination unit 24 ′ that determines the presence / absence of abnormality of the product, and an apparatus abnormality determination unit 24 ″ that determines the presence / absence of abnormality of the process device itself. The units 24 ′ and 24 ″ can operate independently and simultaneously, so that the abnormality determination unit 24 can perform a plurality of abnormality determinations such as process result abnormality determination (product abnormality determination) and apparatus state abnormality determination. Can be done at the same time.

  As shown in FIG. 6, the process result abnormality determination unit 24 ′ includes an abnormality analysis unit 24′a, an abnormal process data storage unit 24′b, an abnormality notification unit 24′c, and a determination result display unit 24′d. And a determination result storage unit 24′e. The abnormality analysis unit 24 ′ a uses the abnormality detection factor analysis rule stored in the abnormality detection factor analysis rule data storage unit 26 to perform abnormality determination according to the process feature amount read from the process feature amount data storage unit 23.

  As shown in FIG. 7, the apparatus abnormality determination unit 24 ″ includes an abnormality analysis unit 24 ″ a, an abnormal process data storage unit 24 ″ b, an abnormality notification unit 24 ″ c, a determination result display unit 24 ″ d, And a determination result storage unit 24 ″ e. The abnormality analysis unit 24 ″ a uses the abnormality detection factor analysis rule stored in the abnormality detection factor analysis rule data storage unit 26 to perform abnormality determination according to the process feature amount read from the process feature amount data storage unit 23.

  6 and 7, the process result abnormality determination unit 24 ′ and the apparatus abnormality determination unit 24 ″ have the same basic configuration, and the abnormality analysis units 24′a and 24 ″. The abnormality detection factor analysis rule used in a is different.

  The abnormality determination performed by the abnormality analysis unit 24′a of the process result abnormality determination unit 24 ′ is both the presence / absence of abnormality of the product generated as a result of executing the process and the abnormality factor analysis. The abnormality determination executed by the abnormality analysis unit 24 ″ a of the apparatus abnormality determination unit 24 ″ is both the presence / absence of abnormality of the process device (process state) and the abnormality factor analysis. A specific processing algorithm of each abnormality analysis unit 24′a, 24 ″ a will be described later.

  The abnormal process data storage units 24 ′ b and 24 ″ b store process data on wafers determined to be abnormal when the abnormality analysis units 24 ′ a and 24 ″ a detect abnormalities. Is stored in the abnormal process data storage unit 27 as abnormal process data. At this time, the abnormality determination result (value of y) may be associated and registered.

  The abnormality notification unit 24′c, 24 ″ c outputs an abnormality message to the specified abnormality display device 2 when an abnormality is detected by the abnormality analysis unit 24′a, 24 ″ a (abnormality notification). . The abnormality message to be output is stored in the abnormality detection factor analysis rule data storage unit 26. In addition, the abnormality notification unit 24'c, 24 "c also has a function of outputting an abnormality message by a designated method to a designated abnormality notification destination. As an example, the abnormality output unit 24'c, 24" is provided. c sends a mail to the designated address. The abnormality message to be output is stored in the abnormality detection factor analysis rule data storage unit 26. The notification destination can be specified for each rule.

  The determination result display units 24′d and 24 ″ d display a specific abnormality determination result on the abnormality display device 2 in a designated display form. When abnormality factor analysis is performed, details such as a contribution rate are displayed. Data is also output.

The determination result storage units 24′d and 24 ″ d store the result of abnormality determination in the abnormality analysis units 24′a and 24 ″ a in the determination result data storage unit 28 as determination result data. That is, the result of the abnormality determination is stored together with the PLS predicted value, the predicted value contribution rate, the Q, T ² statistic, and the like, and can be searched from the abnormality display device or the like. This determination result data may be stored for all determination results, or may be stored only when determined to be abnormal.

  Specific processing functions of the abnormality analysis unit 24'a of the process result abnormality determination unit 24 'are as shown in the flowcharts of FIGS. The process data editing unit 22 collects process data for one sheet of product (S1), and calculates a process feature amount from the process data (S2). The calculated process feature quantity is stored in the process feature quantity data storage unit 23.

  The abnormality analysis unit 24'a accesses the process feature value data storage unit 23, extracts process feature value data for one sheet using one product ID as a key, and acquires the recipe ID. Then, the abnormality analysis unit 24′a accesses the abnormality detection factor analysis rule data storage unit 26 and acquires the abnormality detection factor analysis rule corresponding to the acquired recipe ID (S3).

The abnormality analysis unit 24′a calculates a process result predicted value (y value), a predicted value contribution rate, a Q statistic, and a T ² statistic based on the acquired process feature amount and the abnormality detection factor analysis rule ( S4). The abnormality analysis unit 24'a determines whether or not Q is in the normal range (S5). If the Q is within the normal range, the count value of the Q abnormality counter is reset to 0 (S6). The count value of the Q abnormality counter is incremented by 1 (S7). Similarly fault analysis unit 24'a is, T ² is determined whether the normal range (S8), the count value of the abnormality counter T ² are the case of the normal range is reset to 0 (S9), normal when out of range increments the count value of the abnormality counter T ² (S10).

Fault analysis unit 24'a the count value of the abnormality counter of Q and T ² is determined whether both of them are 0 (S11), when none of the 0 determines the process state to a normal state ( S12) If at least one is not 0, the process state is determined as an abnormal state (S13).

  When the process state is normal, the abnormality analysis unit 24′a determines whether or not the process processing result prediction value (PLS prediction value or the like) is in a normal range (S14). The count value of the abnormal counter of the value (PLS predicted value etc.) is reset to 0 (S15), and if it is out of the normal range, the count value of the abnormal counter of the process processing result predicted value (PLS predicted value etc.) is incremented by 1. (S16).

After executing the processing step S17 or S18, the abnormality analysis unit 24a determines whether or not the count value of the abnormality counter of the process processing result predicted value (PLS predicted value or the like) is less than the specified number (S17). In the above case, the process result predicted value, the predicted value contribution rate, and the Q and T ² statistics are notified (S18).

  As described above, the abnormality analysis unit 24′a of the process result abnormality determination unit 24 ′ determines whether or not there is an abnormality in the process state. When the process state such as an abnormality has occurred in the device that executes the process, it is impossible to predict the correct process result, and therefore, the process is performed by executing steps S4 to S13. Try to eliminate the case. As a result, the process device operates normally, and it is guaranteed that the collected process data accurately reflects the process state in the process device. Then, the process processing result by PLS is predicted and the result is notified. can do.

Note that the processing steps S12 and S13 are not particularly required if the above-mentioned “accurate reflection” is guaranteed. Also, if the process state is abnormal, the abnormality analysis unit 24'a determines whether less than the count value specified number of abnormality counter of Q or T ^2, if the above specified number, the process state (critical) A function of notifying the abnormality with the contribution rate may be provided.

  The notification of abnormality notifies the abnormality according to the abnormality notification information corresponding to the preset determination condition. Specifically, the abnormality output unit 24′c notifies the abnormality by outputting a message to a preset abnormality display device 2 or sending a mail to a preset abnormality notification destination. . In addition to the abnormality display information stored in the abnormality detection factor analysis rule data storage unit 26 and the recipe ID, occurrence date / time information and an abnormality notification ID are added to the contents to be notified.

  The specific processing functions of the abnormality analysis unit 24 ″ a of the device abnormality determination unit 24 ″ are as shown in the flowcharts of FIGS. The process data editing unit 22 collects process data for one sheet of product (S21), and calculates a process feature amount from the process data (S22). The calculated process feature quantity is stored in the process feature quantity data storage unit 23.

  The abnormality analysis unit 24 ″ a accesses the process feature amount data storage unit 23, extracts process feature amount data for one sheet using one product ID as a key, and acquires the recipe ID. The analysis unit 24 ″ a accesses the abnormality detection factor analysis rule data storage unit 26 and acquires the abnormality detection factor analysis rule corresponding to the acquired recipe ID (S23).

The abnormality analysis unit 24 ″ a calculates a Q statistic and a T ² statistic based on the acquired process feature amount and the abnormality detection factor analysis rule (S24). The abnormality analysis unit 24 ″ a determines that the Q is in a normal range. (S25), if it is within the normal range, the count value of the Q abnormality counter is reset to 0 (S26). If it is outside the normal range, the count value of the Q abnormality counter is incremented by one. (S27). Similarly the fault analysis section 24 "a is, T ² is determined whether the normal range (S8), the count value of the abnormality counter T ² are the case of the normal range is reset to 0 (S9), normal when out of range increments the count value of the abnormality counter T ² (S30).

Fault analysis unit 24 "a count value of the abnormality counter of Q and T ² is determined whether both of them are 0 (S31), if any of 0, it is determined process state and a normal state ( S32) If at least one is not 0, the process state is determined as an abnormal state (S33).

Then, if the process state is abnormal, the abnormality determination section 24 "a count value of the abnormality counter Q or T ² is determined whether the below specified number (S34), if more than the specified number of times, the process conditions (Severe) Notification is made together with the contribution rate as an abnormality (S35).

  The notification of abnormality notifies the abnormality according to the abnormality notification information corresponding to the preset determination condition. Specifically, the abnormality output unit 24′c outputs a message to the preset abnormality display device 2, or notifies the preset abnormality notification destination by mail transmission or the like. In addition to the abnormality display information stored in the abnormality detection factor analysis rule data storage unit 26 and the recipe ID, occurrence date / time information and an abnormality notification ID are added to the contents to be notified.

  An example of display displayed on the display device of the abnormality display device 2 based on the abnormality notification is shown in FIG. The display screen of FIG. 12 shows an example of the device state abnormality determination result screen. When the abnormality display device 2 receives the abnormality notification and the determination result of the process device (process state), the abnormality display device 2 sets the message in the display area of each state to “abnormal”. Further, the history of the received abnormality notification is stored and the history information is also displayed. FIG. 12 shows an example of the display screen for the process device, but the process result abnormality can be displayed in the same manner.

  Furthermore, the abnormality display device 2 is not limited to the abnormality monitor and the history information display showing the current state shown in FIG. 12, but may display a predicted value trend as shown in FIG. 13 or as shown in FIG. The predicted value high contribution factor factor is displayed, the factor trend of the high contribution factor is displayed as shown in FIG. 15, or the time series raw data of the high contribution factor factor is displayed as shown in FIG. Various display forms can be taken. 13 to 16 show examples of process result abnormality display.

It is a figure which shows an example of the manufacturing system containing the abnormality detection apparatus which is suitable embodiment of this invention. It is a figure which shows an example of the internal structure of an abnormality detection apparatus. It is a figure which shows an example of the data structure of the various data which an abnormality detection apparatus processes. It is a figure which shows an example of the data structure of the rule data stored in an abnormality detection factor analysis rule data storage part. It is a flowchart explaining the function of an abnormality detection factor analysis rule edit part. It is a figure which shows an example of the internal structure of a process result abnormality determination part. It is a figure which shows an example of the internal structure of an apparatus abnormality determination part. It is a flowchart explaining the function of the abnormality analysis part of a process result abnormality determination part. It is a flowchart explaining the function of the abnormality analysis part of a process result abnormality determination part. It is a flowchart explaining the function of the abnormality analysis part of an apparatus abnormality determination part. It is a flowchart explaining the function of the abnormality analysis part of an apparatus abnormality determination part. It is a figure which shows an example of the information displayed on an abnormality display apparatus. It is a figure which shows an example of the information displayed on an abnormality display apparatus. It is a figure which shows an example of the information displayed on an abnormality display apparatus. It is a figure which shows an example of the information displayed on an abnormality display apparatus. It is a figure which shows an example of the information displayed on an abnormality display apparatus.

Explanation of symbols

20 abnormality detection device 21 process data storage unit 22 process data editing unit 23 process feature data storage unit 24 abnormality determination unit 24 ′ process result abnormality determination unit 24 ″ device abnormality determination unit 24′a, 24 ″ a abnormality analysis unit 24 ′ b, 24 ″ b Abnormal process data storage unit 24′c, 24 ″ c Abnormal notification unit 24′d, 24 ″ d Determination result output unit 24′e, 24 ″ e Determination result storage unit 25 Abnormality detection factor analysis rule editing unit 26 Abnormality detection factor analysis rule data storage unit 27 Abnormal process data storage unit 28 Judgment result data storage unit

Claims (7)

In a manufacturing system composed of one or a plurality of manufacturing apparatuses, an abnormality detection apparatus that detects a process abnormality based on process data obtained during process execution,
Process data storage means for storing the time-series process data;
Process data editing means for extracting process feature values from the process data stored in the process data storage means;
An anomaly detection rule data storage means for storing an anomaly detection rule for detecting an anomaly in the manufacturing system from a process feature;
A plurality of abnormality determination means for determining presence / absence of abnormality related to the process from the process feature amount by the abnormality detection rule,
With
The plurality of abnormality determination means are configured to be executable simultaneously,
At least one of the plurality of abnormality determination means shall determine the presence or absence of abnormality by an abnormality detection rule that requires an objective variable,
Another one of the plurality of abnormality determination means is configured to determine whether or not there is an abnormality based on an abnormality detection rule that does not require an objective variable.   2. The process abnormality detection apparatus according to claim 1, wherein the abnormality detection rule that requires the objective variable is for determining whether or not there is an abnormality in a product obtained by executing the process.   The process abnormality detection apparatus according to claim 1, wherein the abnormality detection rule that requires the objective variable includes a rule that uses a partial least square regression model as a prediction model of a process execution result.   The process abnormality detection apparatus according to claim 1, wherein the abnormality detection rule that does not require the objective variable is for determining whether or not there is an abnormality in the manufacturing apparatus for executing the process. The process according to any one of claims 1 to 4, wherein the anomaly detection rule that does not require the objective variable includes a rule that uses a Q statistic and / or a T ² statistic by principal component analysis. Anomaly detection device. In a manufacturing system composed of one or a plurality of manufacturing apparatuses, a method in a process abnormality detecting apparatus for detecting a process abnormality based on process data obtained during process execution,
Acquiring time series process data and storing it in the process data storage means;
Extracting a process feature amount from the process data stored in the process data storage means;
An abnormality determination step for determining whether or not there is an abnormality related to the process from the extracted process feature value;
Is to execute
In the abnormality determination step, a process for determining the presence / absence of an abnormality by an abnormality detection rule that requires an objective variable and a process for determining the presence / absence of an abnormality by an abnormality detection rule that does not require an objective variable are simultaneously executed. Process abnormality detection method characterized by the above. Computer
Process data editing means for extracting process feature values from time-series process data stored in the process data storage means;
An abnormality determination means for determining presence / absence of an abnormality related to the process from the process feature amount by an abnormality detection rule that requires an objective variable;
Another abnormality determining means that is executed at the same time as the abnormality determining means and determines whether or not there is an abnormality related to the process from the process feature amount by an abnormality detection rule that does not require an objective variable.
Program to function as.

Images